How Authoritarian Governments Decide Who Emigrates: Evidence from East Germany

Abstract

Most autocracies restrict emigration yet still allow some citizens to exit. How do these regimes decide who can leave? We argue that many autocracies strategically target anti-regime actors for emigration, thereby crafting a more loyal population without the drawbacks of persistent co-optation or repression. However, this generates problematic incentives for citizens to join opposition activity to secure exit. In response, autocracies simultaneously punish dissidents for attempting to emigrate, screening out all but the most determined opponents. To test our theory, we examine an original data set coded from over 20,000 pages of declassified emigration applications from East Germany's state archives. In the first individual-level test of an autocracy's emigration decisions, we find that active opposition promoted emigration approval but also punishment for applying. Pensioners were also more likely to secure exit, and professionals were less likely. Our results shed light on global migration's political sources and an overlooked strategy of autocratic resilience.

In 2011, as part of a crackdown on government critics, China jailed artist Ai Weiwei and confiscated his passport. After his release, Ai placed a bouquet on a bike outside his apartment every morning to protest continued restrictions on his travel. In 2015, with little warning, the regime reversed course, returning Ai's passport and allowing him free travel abroad.² China's treatment of Ai is emblematic of how most autocracies manage emigration, neither respecting complete freedom of exit nor enforcing total closure. Instead, most modern autocracies employ a selective emigration policy that allows some exit but limits the right to specific individuals or groups. This includes many of the most repressive regimes, from Cuba to the Soviet Union and the Eastern Bloc to modern Russia and Venezuela. How do autocracies strategically manipulate emigration to their benefit? Specifically, how do they decide which citizens are allowed to exit and which are forced to stay?

Scholarship on emigration policy in autocracies has focused on the general openness of exit.³ In this article we shift the focus to who regimes allow to emigrate. A central motive for freer exit is that emigration can act as a “safety valve” that allows political opponents to leave, reducing opposition at home.⁴ However, this comes with a dangerous sacrifice of control if the regime loses the ability to retain citizens vital to the economy and security. We examine a widespread approach in which regimes deliberately select opponents for legal emigration while restricting the right of others they want to keep, which we call a targeted safety valve strategy. This turns the famous Hirschman framework of individuals choosing between exit and voice on its head—autocrats also have agency in targeting their opponents for easy exit.⁵ We argue that this strategy has several advantages over co-optation or repression alone.

Although many qualitative accounts describe the targeted safety valve in action, there exists no direct empirical test of opposition targeting nor a theoretical explanation of how regimes handle the major incentive problems that arise. In particular, citizens can become motivated to express opposition to secure exit. We argue that regimes using this strategy to select out opponents will allow exit in combination with significant punishment. This serves as a screening mechanism that ensures only the most motivated opponents leave. Regimes also tailor their strategy to reduce the economic and security costs from the loss of productive citizens.

To test how autocracies decide who exits and who stays, we examine a random sample of declassified emigration applications from East Germany. Whereas previous studies have examined the characteristics of successful emigrants, we also need to understand who applied and got denied. To our knowledge, this is the first large-scale empirical analysis of an autocracy's individual-level decisions on emigration.

Although East Germany is often imagined as closed between the Berlin Wall's construction in 1961 and the mass emigration crisis of late 1989, it authorized an average of 25,000 citizens to emigrate each year from 1962 to 1988.⁶ As we discuss, the regime formulated these individual decisions with great care, making it an ideal case to tease out an underlying logic. The hierarchical structure of the decision process also helps mitigate sample selection bias and other forms of confounding. Moreover, among regimes with application data available, East Germany's unusually extensive data gathering for its time makes it arguably the best parallel to contemporary autocracies and their uses of digital and surveillance technology.

Our empirical analysis of more than 500 randomly selected application files (consisting of more than 20,000 pages) shows that East Germany was significantly more likely to allow emigration if the applicant engaged in active opposition, demonstrating the targeted safety valve strategy. This effect is robust across multiple testing variations, including panel setups, and is larger if the applicant used more threatening forms or numerous types of opposition. Opposition also increased the likelihood of punishment (measured by surveillance and imprisonment) in response to the application, illustrating the screening mechanism that coexists with the safety valve. We also find evidence that East Germany resisted letting economically valuable citizens leave, capturing the trade-offs presented by citizen exit.

A deeper understanding of how autocracies select individuals for exit is important for two reasons. First, it illuminates a hidden logic behind a significant source of global migration. As of 2019, about 125 million of the world's 270 million migrants emigrated from an autocracy,⁷ with the vast majority exiting through legal channels.⁸ Although not all autocracies select emigrants in the same way, there are numerous politically significant cases of autocracies allowing out or pushing out citizens they regard as undesirable, from Cuba's Mariel boatlift to the Soviet Union's policy toward Jews and other minorities starting in the 1970s to modern Russia and China.⁹ We examine the decision-making process underlying both the level of emigration and the specific individuals exiting, with substantial economic and political implications for sending and receiving countries.¹⁰

Second, strategic emigration is an underexamined tool of autocratic resilience. The targeted safety valve strategy reduces opposition without the negative consequences of indefinitely imprisoning or killing opponents. Moreover, it can help expose hidden opposition, since applicants approach the state to apply and thus reveal their dissatisfaction. The disproportionate exit of political opponents has had significant stabilizing effects on regimes like Venezuela, Russia, Zimbabwe, and Hungary.¹¹ This effect should only accelerate in an increasingly mobile world. In essence, we can understand emigration policy as a state's way of crafting its domestic citizenry to minimize opposition and maximize economic returns.

Emigration Policy in Autocracies

Patterns of Emigration Policy

Migration is a significant economic and political force in the modern world, with more than 270 million people currently living abroad. Emigration is especially salient for autocratic sending countries, given strong evidence that autocratic emigration predicts the spread of democratic norms, political contestation, and regime change.¹² A growing body of research examines how autocracies strategically determine the freedom of exit to bolster regime survival.¹³

Autocrats need to weigh several costs and benefits of freer emigration policy.¹⁴ Costs include greater citizen leverage over their rulers,¹⁵ the threat of ideological influence from abroad,¹⁶ and the potential for brain drain, in which highly educated or economically valuable citizens leave.¹⁷ On the positive side, emigration brings economic benefits like remittances, increased trade and investment, and skills upgrading among returnees.¹⁸ In addition, freer exit can act as a safety valve by allowing regime opponents, unallied ethnic groups, and the economically frustrated to leave,¹⁹ which in turn reduces anti-regime protest and civil conflict.²⁰

Figure 1 shows the level of emigration freedom across the world every twenty years from 1960 to 2020. Whereas democracies generally allow free exit, autocracies display a wide range of policies. Although there has been a slow trend toward greater emigration freedom, most autocracies (about 82%) remain somewhere between wholly restricted (e.g., North Korea and Eritrea) and entirely free.²¹ In these “partly restricted” regimes, states retain control over which individuals and groups can exit. We refer to this as a policy of selective emigration.

FIGURE 1. Emigration Freedom in the World, 1960 to 2020

Countries use different tactics to select who gets to emigrate. Many states use exit visas that require individuals to apply for permission to leave the country, while others limit who can obtain a passport, or implement no-travel lists. On the other end of the spectrum, states can use the visas granted under bilateral labor agreements to target specific communities for exit.²² Figure 2 shows the fraction of autocracies using exit visas from 1973 to 2020 and the fraction using no-travel lists and other restrictions from 2008 to 2020. Over the last fifteen years, about two-thirds of autocracies have used exit visas, no-travel lists, or other restrictions (typically passport controls) to regulate exit. For instance, as of 2020, Bhutan, China, Equatorial Guinea, Myanmar, Singapore, and Thailand all restricted passport access.

FIGURE 2. Fraction of Autocracies Requiring an Exit Visa, 1973 to 2020

Explaining Selective Emigration

The prevalence of selective emigration in autocracies points to a significant open question in the migration literature: How do autocracies determine who is allowed to exit? We focus on whether autocracies deliberately target opponents for emigration, which we call the targeted safety valve strategy. By facilitating the emigration of political opponents, regimes can ensure a more loyal population at home. Several qualitative accounts posit that autocracies employ this strategy. For instance, the British allowed anti-colonial Irish dissenters to emigrate; the Soviets allowed hundreds of thousands of Jews to move abroad;²³ Bulgaria, Poland, and Yugoslavia also allowed ethnic minorities to emigrate;²⁴ and in the 1940s, Mexico concentrated exit permits under the Bracero program in regions with high opposition-party membership and civil unrest.²⁵

However, the strategic soundness of the targeted safety valve is disputed. Sending opponents abroad risks creating an influential anti-regime diaspora beyond the autocracy's control.²⁶ Instead of seeding this diaspora with political opponents, regimes may want to keep their enemies close. Further, there is a neglected open puzzle regarding the targeted safety valve: How do autocracies manage the problem that it rewards opposition? We posit a screening mechanism in which emigration applicants face significant punishment before being allowed to exit, ensuring that only the most dedicated opponents complete the process.

Besides these theoretical disputes, the qualitative evidence that autocracies employ the targeted safety valve is limited. These accounts typically rely on observations that political opponents and oppressed groups are disproportionately represented among emigrants. However, it is often unclear whether this is because opponents are more likely to want to leave or because the regime actively encourages it.

Theory on Selective Emigration

How do autocrats select individuals and groups for exit? we outline a targeted safety valve strategy that provides an alternative to pure repression or co-optation. We argue that autocracies implement this strategy while balancing the economic and security costs of letting specific individuals leave. To clarify our scope conditions, our theory applies to autocracies lying between total closure and free exit, allowing some citizens to exit. We also assume the autocracy has sufficient state capacity to control emigration through laws or bureaucratic scrutiny.²⁷ Although we test our theory in East Germany, our hypotheses are not limited to it, and we explain later how the balance of motives might vary across cases.

Citizens’ Desire for Exit

For exit to be a regime strategy, some citizens who oppose the government must want to leave. Citizens vary in their opposition to the regime for normative and instrumental reasons, which will affect their desire to emigrate. Regime supporters, including insiders and those who benefit from regime policies, are happy with the status quo and generally favor staying. They are also more likely to have opportunities to voice their concerns about the regime that may result in change rather than punishment. For everyone else, as dissatisfaction increases, citizens become increasingly motivated to decide between “exit” and “voice” to improve their well-being,²⁸ depending on the costs of each. The cost of exit includes factors the regime has control over (such as the cost of a passport²⁹ and any punishment for applying for permission to exit) and factors it has less control over (such as the cost of transportation or the ability to get a visa for another country).³⁰ If the costs for exit are high enough, even the most determined to emigrate will not apply.

It may seem surprising that some regime opponents want to leave, but winning positive domestic change through “voice” comes with significant personal risk and faces a major collective-action problem. Successful mass mobilization is often infeasible without major regime failures like stolen elections or violent internal disruption.³¹ Emigration is particularly attractive to members of what one can call the motivated middle class. In contrast to the acquiescent, state-dependent bureaucratic class,³² these are high-skill individuals who are unhappy with their current situation and willing to act to change it. Although they are natural supporters of opposition groups and democracy if forced to stay, they have the skills and resources to pursue a better life elsewhere. Especially when facing an entrenched dictatorship, “exit” rather than “voice” is frequently the more realistic and direct route to such positive change for regime opponents.

The Regime's Strategy

We assume that an autocratic leader's primary goal is to stay in power, which requires managing threats from rival elites and opposing citizens.³³ The modern autocracy literature emphasizes two primary strategies for managing opponents: co-optation and repression.³⁴

We augment this framework. In addition to co-optation and repression, dictators can select out opponents for emigration: a targeted safety valve strategy. Research in the “exit versus voice” framework has long posited that exit reduces the ranks of dissidents exercising voice, but typically assumes exit costs are the same for all citizens³⁵ or are not determined by the state.³⁶ Instead, we endogenize the autocrat's choice to expand or restrict exit opportunities for specific populations. By systematically selecting out active or potential dissidents, regimes can craft a more loyal population and hamper opposition organization and leadership. Free exit may disproportionately attract anti-regime citizens but also risks losing the economically valuable along with opponents. The ideal is to let out only the citizens whose presence challenges regime survival. When using selective emigration, regimes need to balance several competing objectives.

Benefits of the Targeted Safety Valve Strategy

The targeted safety valve has significant advantages over repression and co-optation by themselves. First, it helps solve information problems. Identifying dissatisfied citizens in repressive autocracies is a considerable challenge, sometimes called “the dictator's dilemma,”³⁷ because individuals have incentives to hide their opposition until they sense regime weakness.³⁸ By requiring citizens to apply for exit, selective emigration drives unhappy citizens to engage with the state and reveal their unhappiness. Co-optation strategies can also invite expressions of dissatisfaction to win concessions, but the dictator will be uncertain whether they are offering more than needed for loyalty. In contrast, only citizens who genuinely want to leave will complete the emigration process.

Second, opposition emigration is a formal, voluntary process that often costs much less than co-optation or repression. Co-optation can entail extensive and indefinite payoffs, major policy concessions, or institutional change with uncertain consequences. Further, dictators face a budget constraint: expanding who benefits from the regime leaves less for other regime insiders. Repression requires costly investment in coercive capacity, impedes economic development, and is increasingly met with international punishment.³⁹ It also intensifies principal–agent problems by strengthening security apparatuses that might betray the dictator.⁴⁰

In contrast, selective emigration requires neither payoffs nor a buildup of state capacity outside a (relatively) small, unarmed bureaucracy to process applications. Autocratic leaders may also gain international goodwill or monetary compensation when they allow opponents to leave. The 1974 Jackson–Vanick amendment in the United States tied trade relations with socialist countries to their liberalization of emigration.⁴¹ Indeed, autocratic rulers seem to realize the lower cost of selective emigration: In recent years, the Kremlin “appears to be betting that forcing high-profile critics out of the country is less of a headache than imprisoning them.”⁴²

Third, the targeted safety valve is ideal for sapping opposition strength, as it can select out the most motivated opponents, potential leaders, and critical members of social networks. Seeing their allies escape abroad is often disruptive and demoralizing for the opposition activists left behind. In contrast, institutional co-optation gives the opposition political visibility and opportunities to increase domestic support.⁴³ In repressive regimes, attempts to “decapitate” opposition groups by killing or jailing leaders risks mobilizing the opposition and inciting backlash across society, including among regime supporters.⁴⁴

Given these advantages, it is unsurprising to find numerous examples of autocrats targeting political opponents and hostile out-groups for emigration. Among the most politically significant examples is the 1980 Mariel boatlift, during which the Castro regime selected for exit several thousand current and former political prisoners and about 10,000 protesters who had occupied the Peruvian embassy.⁴⁵ Despite the seeming chaos, the regime maintained tight control over who was allowed to leave, with a stream of “undesirables” targeted by the government “to alleviate Cuba's domestic, economic, and social pressures.”⁴⁶ Average citizens could apply to exit only if a family member from the United States picked them up, limiting the outflow to those with close ties to the anti-Castro diaspora.⁴⁷ Even then, state agents scrutinized applicants for weeks.⁴⁸

The 1970s Soviet policy targeting Jews for emigration aimed “to minimise Jewish nationalist activity and protest without recourse to mass arrests and deportations.”⁴⁹ Soviet dissident Zhores Medvedev argued in 1978 that “the state now manipulates emigration opportunities for its own convenience, often just to rid itself of dissidents, the old and the useless.”⁵⁰ More recently, convinced that “not enough dissidents were allowed to leave the country in the Soviet era,” Russian leaders have increasingly pushed opposition leaders abroad, where they “are easy to paint as traitors in cahoots with the West.”⁵¹ Qualitative scholars have also described this strategy in Mexico, Morocco, Iran, and elsewhere.⁵² However, rigorous individual-level evidence for this targeting has remained elusive.

A closely related strategy is forced exit, in which regimes exile opponents instead of relying on voluntary emigration.⁵³ The practice of deporting dissidents has been widespread, including in Chile,⁵⁴ the Soviet Union,⁵⁵ and Ivory Coast, where in 1958 Félix Houphouët-Boigny declared that anyone voting the wrong way in a referendum would have twenty-four hours to leave the country.⁵⁶ History has also seen many group-level examples, from the exile of Jews from England and Spain to Idi Amin's 1972 expulsion of Uganda's South Asian minority.

We expect forced exit to follow a targeting logic similar to selective emigration. However, like mass imprisonment, forced exit has much higher costs in coercive needs and information. Further, creating flows of forced exiles or refugees can damage a state's international reputation. In extreme cases, it can lead to sanctions under international law, as with the mass exodus of Rohingya from Myanmar. As a result, forced exit is not nearly as common today as legal emigration.

In sum, the targeted safety valve is a highly effective strategy for draining the opposition without the drawbacks of pure co-optation or repression. This leads to our first hypothesis:

H1 Individuals who have expressed active opposition to the regime are more likely to be allowed to emigrate.

Potential Cost 1: Adverse Selection

A potential problem for the autocrat is that, simply put, rewarding opposition gets you more of it. Despite the appeal of filtering out opponents through emigration, public knowledge of this strategy will create perverse incentives for many citizens. First, it lessens the risk of opposition by providing an easy way out if movements are unsuccessful, encouraging wary political dissidents to become active. Indeed, Hirschman theorizes that whereas the exercise of exit atrophies voice, the opportunity for exit can increase voice.⁵⁷ Second, the strategy encourages noisy opposition among average citizens who want to leave but would otherwise remain quiet. By inducing opposition, the strategy would quickly lose its informational value and drive out many acquiescent citizens the regime would prefer to keep. The autocrat's goal should be to select out only those most likely to engage in opposition.

We argue that dictators frequently use a screening technique that punishes opponents when they apply for emigration to deter the less determined. Successful applicants may weather years of threats, grueling interrogations, and economic penalties before leaving. In East Germany, the regime's immediate response was almost always pressure to withdraw the application, combined with promised rewards such as better housing or work assignments.⁵⁸ As a first cut, this sifted out the applicants with minor grievances who could be bought off.⁵⁹ Anti-regime actors who maintained their applications were then usually met with retaliation—placed under surveillance, discredited within their social networks, fired from their jobs and forced into manual labor, and sometimes imprisoned.⁶⁰

Similarly, the Cuban regime ordered citizens and state agents to carry out “acts of repudiation” against emigration applicants during the Mariel boatlift.⁶¹ Applicants were beaten by mobs, egged, and attacked with police dogs.⁶² One childhood witness to Mariel recalls his teacher ordering the class to vandalize the homes of would-be emigrants after school.⁶³ The Soviet Union targeted Jews who expressed a desire to leave, including firing them from their jobs, expelling their children from university, and imprisoning them.⁶⁴ Thus, repression, co-optation, and the targeted safety valve often function as complements rather than discrete alternatives.

By imposing immediate and sustained costs on applicants, dictators obtain credible signals of their resolve. Only the most determined and dissatisfied opponents will endure the punishment. Since these citizens are the most likely to continue organizing against the regime if they stay, regardless of attempts at repression or co-optation, the best alternative is to let them leave. In contrast, the moderately dissatisfied will be deterred from applying or convinced to retreat. In the language of economics, this is an example of screening to solve the adverse selection problem generated by asymmetric information (in this case, opposition intensity).

Another benefit of this strategy is that the severity of punishment can be calibrated to obtain the total amount of emigration desired by the state without abandoning the selection mechanism. Punishing individuals too harshly for their attempt to exit could increase hostility to the regime and deter even the most determined opponents from applying for emigration. Further, regimes can sensibly target punishment at existing opponents and not, say, those the regime wants to leave for economic reasons. For example, pensioners who apply for emigration can be allowed easy exit, while working-age adults are punished. In sum, by making exit costly, regimes can credibly reveal opposition intensity and select those they want to emigrate.

H2 Emigration applicants who have expressed active opposition to the regime are more likely to face punishment in response to the application.

Potential Cost 2: Organizing Abroad

A second consideration is that individuals can organize abroad. Opposition members may be able to learn more about their home state's repression, freely organize, and communicate with friends and family back home about their better life abroad or their government's activities. They can reach a wider audience through traditional media (like Radio Free Europe / Radio Liberty) and social media. In an extreme case, the United States armed a Cuban exile group, Brigade 2506, to attempt to overthrow the Castro regime in the Bay of Pigs invasion. More often, democratic governments fund these groups through democracy-promoting programs.

Although regime opponents can organize abroad, this typically presents less of a challenge than if the same actors organized at home.⁶⁵ Actual regime overthrow, as opposed to information dissemination, becomes more difficult when the opposition is not physically present. For over sixty years, the Cuban diaspora in the United States has effectively organized and obtained help from the American government in its fight against the Castro regime. Yet the regime has remained in power.

Moving abroad also does not put dissidents beyond regimes’ reach. Many autocratic governments have effectively controlled opposition diasporas through infiltration, surveillance, and harassment.⁶⁶ Freedom House recently documented 735 incidents of transnational repression between 2014 and 2021, including numerous examples of the Russian, Turkish, and Thai authorities assisting in the detention and deportation of foreign dissidents.⁶⁷ This even extends to liberal democracies, with the Friendship Societies of Algeria and Morocco and the Chinese Students and Scholars Association surveilling and harassing dissidents in Europe and the United States.⁶⁸ East Germany similarly used informants and spies to keep tabs on its diaspora.⁶⁹ In addition, autocracies try to delegitimize opposition abroad. Mao called emigrants “traitors,” as did the PRI in Mexico and, more recently, President Putin of Russia.⁷⁰

Potential Cost 3: Losing Productive Citizens

In addition to managing opposition, autocracies must weigh emigration's economic and security impacts. Allowing individuals to leave implies they will not directly contribute to the domestic economy but also will not consume state resources. Depending on the individual, this can impose high costs on the regime or provide a benefit.

A potential economic cost of emigration is the loss of human capital. Although low-skill emigration can shrink the labor force in industry and agriculture, making production more expensive, the loss of professional labor (brain drain) is typically seen as more problematic since it may impede technical advancement and economic growth. It can also imperil the provision of public goods like defense, health care, or education. Syria, for instance, prevents engineers from exiting until they work for the state for five years.⁷¹ Similarly, in our East Germany data, we find state agents denying emigration to medical professionals, noting “the loss of a doctor or nurse due to emigration infringes in major ways on the quality of life of sick citizens, which cannot be justified” (application notes, translated by the authors). In general, authoritarian states should be reluctant to allow professionals to leave.⁷²

On the other hand, regimes can benefit if those who leave are unproductive and net consumers of social welfare. During the Mariel boatlift, Cuba sent off a “number of criminals, the mentally ill, and the chronically infirm,”⁷³ intending to pass their social welfare costs on to the United States.⁷⁴ In other cases, the state can win immediate economic rewards from emigration. After the Jackson–Vanick amendment's passage, observers noted a correlation between American trade with the Soviet Union and the emigration of Soviet Jews.⁷⁵ West Germany sometimes offered ransom payments for East German emigrants, especially political prisoners. However, this affected a relatively small number of cases. Pfaff estimates that about 5 percent of successful emigrations included a ransom paid by West Germany, usually routed through a network of churches.⁷⁶ Similarly, we find references to payments in about 2 percent of our application sample.⁷⁷

The economic balance will also depend on the emigrants’ likely activities abroad and their potential to return home. Migrants, including forcibly displaced individuals, typically remit money back home and foster links to the global economy.⁷⁸ For many developing autocracies, these links can be an essential source of capital and foreign currency. In East Germany, however, the chance of migrants returning and the domestic economic links were extremely limited after the Berlin Wall's construction.⁷⁹ Thus, while the exact balance will differ by regime, our core argument is that autocracies weigh economic returns in deciding who can emigrate. This leads to our final hypothesis:

H3 Individuals with high economic or security value—including professionals, the highly educated, and those in strategic industries like defense and medicine—will be less likely to be allowed to emigrate than those deemed economically unproductive or dependent on the social welfare system.

Emigration from East Germany

Many autocracies use selective emigration to control exit (Figure 2). We test our theory using emigration application decisions in the German Democratic Republic (GDR, or East Germany). Although caution is always warranted when testing theories with microdata from specific countries, several characteristics of East Germany make it an ideal test case.

First, following the Berlin Wall's construction, the regime approached emigration policy with extreme care and strategic consideration. After 1983, exit approvals for regime opponents needed to be signed off by the national head of the Stasi (state security), providing unique insight into inner-circle decision making.

Second, for selective exit to work, individuals the state wants to leave need to gain entry to another state. In the GDR, citizens were automatically granted entry and citizenship in West Germany, meaning we do not need to control for receiving states’ decision-making processes. Third, East Germany's high state capacity gave it substantial control over its border, forcing most citizens to seek state approval for exit. Another benefit of East Germany's state capacity is that its tight control over the bureaucracy makes it highly unlikely that state emigration agents lied in their reports or took bribes. Finally, on the practical side, East Germany has declassified individual emigration documents that are likely the most detailed available from any autocracy.

Emigration Policy as Regime Strategy

The GDR was a communist dictatorship from 1949 to 1990. Throughout its history, the regime repeatedly adjusted its emigration policy to maximize regime stability. Figure 3 shows the number of East Germans leaving each year for West Germany from 1949 through the first nine months of 1989,⁸⁰ alongside the rating by the Varieties of Democracy project of the freedom of foreign movement.⁸¹

FIGURE 3. Emigration from East Germany

After its establishment in 1949, the GDR allowed free exit to purge itself of “fascist remnants.”⁸² It soon faced a severe emigration crisis: 2.7 million East Germans—about 15 percent of its 1949 population, and especially the young and well-educated—left over the next dozen years.⁸³ Technocrats warned that the exodus “imperiled the state's industrial development plan.”⁸⁴ With threats to East Germany's economic future and ideological appeal rising, in 1961 the GDR's leader, Walter Ulbricht, convinced the Soviets to support the construction of the Berlin Wall, along with a sharp restriction on emigration freedom.⁸⁵

From 1961 until its collapse in 1989, East Germany implemented a selective emigration policy that allowed roughly 25,000 citizens out annually.⁸⁶ Although some East Germans tried to leave without state approval, either through border crossings or by overstaying travel visas, authorized emigration dwarfed illegal exits by more than five to one.⁸⁷ Due to the militarization of the East–West border and a highly restrictive approval process for temporary travel for nonpensioners, exit without approval was exceedingly risky.⁸⁸

In 1971, the regime began outlining reasons to potentially allow exit, although it initially kept these reasons secret.⁸⁹ The policy allowed the exit of pensioners and the disabled, as well as citizens seeking family reunification in West Germany or marriage abroad.⁹⁰

In 1976, the regime fully embraced the targeted safety valve strategy, with the interior minister privately declaring, “To minimize the political damage to the GDR, persons with a hostile, negative attitude as well as criminal and asocial elements that persistently pursue emigration…should be allowed to emigrate.”⁹¹ In 1977, a now-declassified directive (Order 6/77) explicitly allowed emigration for citizens displaying anti-regime sentiment to prevent an “immediate and concrete danger to national security” or disruption of the GDR's social order. The directive outlined a range of considerations when approving these “exceptional cases,” including:

(1) the personality of the respective citizen and the information obtained regarding the validity of the motives and reasoning, (2) the type of relations to the individual to whom emigration is supposed to occur, (3) the results of the political-operative measures targeted at disciplining the individual, (4) the extent and character of the applicant's ties back [to the GDR] and their consequences that would emerge through the emigration, (5) the expected reaction to approving emigration by the population or other citizens willing to emigrate.⁹²

We see clear evidence of our screening mechanism here, as agents are asked to weigh whether the anti-regime motives are genuine and whether attempts to punish them have had any success. We also see concern over whether the exit would destabilize the regime through personal contacts or emulation.

Although the state created a targeted emigration system in 1976, applying for emigration was technically illegal until 1983 and often harshly punished.⁹³ For example, public figures such as actors Armin Mueller-Stahl, Eva-Maria Hagen, and Manfred Krug experienced professional retribution before their applications were granted. All applicants were supposed to be put under surveillance.⁹⁴ They also faced the threat of imprisonment: of the 110,000 East Germans imprisoned for political crimes between 1960 and 1989, an estimated 40 to 75 percent were charged in part with trying to emigrate.⁹⁵

The regime, at times, even approved exit applications by imprisoned opponents, including philosopher Rudolf Bahro, writer Juergen Fuchs, and director Freya Klier. Stasi head Erich Mielke explained this choice to fellow Stasi leaders: “Why should they sit here [in our prison]? It would only cause a nuisance, and we have so many political prisoners, and they are eating our food! Why shouldn't they leave?…I have to think economically for our country!”⁹⁶

However, the regime also targeted less prominent dissidents for exit. As one preacher recalls, it was “a thorn in their side that there was a young preacher working with teenagers who was organizing pacifist worship services and not hiding the truth in his sermons, such as about the environmental pollution in the GDR, which he criticized. Such a person they preferred to leave…This is why our application was approved after five months.”⁹⁷

In 1984, the regime experimented with softening emigration control. The main target was political opponents, with tens of thousands of “enemies, criminal elements and incorrigibles” released in the largest exit in two decades.⁹⁸ This softening had the expected effect of weakening the opposition. As Neubert writes, “The mass exodus of 1984 hit opposition groups hard…Such groups were continuously thinned out and therefore at times made temporarily unable to act. This was intended by the security apparatus.”⁹⁹ However, it also had the unintended consequence of quadrupling the number of emigration applications, forcing the regime to dial the policy back and launch a media campaign against emigration.¹⁰⁰

In 1988 and 1989, emigration intersected with the rising protest movement in complex ways. Increasing emigrant numbers revealed large-scale dissatisfaction, while demand for exit became a familiar rallying cry among those left behind.¹⁰¹ Emigration, however, also created bitter divisions within the opposition, as many regarded mass exit as sapping the opposition's strength and unity. Within dissident groups, “the turnover created as people left the country meant there was little familiarity or communication within the group anymore.”¹⁰² A moderate reform group's appeal in a newspaper declared, “Those who leave diminish our hope. We beg you, stay in your homeland, stay with us.”¹⁰³ This weakening effect is in line with the targeted safety valve strategy and was embraced by East Germany's leadership, with party leader Erich Honecker asserting, “I will not shed a single tear for those who want to leave the country.”¹⁰⁴ Indeed, evidence suggests that higher local emigration rates in the 1980s reduced the magnitude of protest in 1989.¹⁰⁵

By the end of 1989, overwhelming internal opposition and a mass outflow of East Germans through socialist neighbors to the West spelled the end of the regime. Tellingly, the regime stuck to its safety valve strategy to the bitter end. On 9 November party leaders formulated a delaying tactic, liberalizing emigration but still requiring a lengthy process of applying for a passport and acquiring Western currency.¹⁰⁶ That night, large crowds approached the Berlin Wall, incorrectly believing that exit had been made free. A commander of one border crossing was told by superiors to “seek out the ‘more aggressive’ people at the checkpoint, note down their names and let them through with a special stamp” to ensure they could not return.¹⁰⁷ Of course, this last-gasp tactic failed, ending the regime and twenty-eight years of selective emigration.

Application Process

Regime leaders crafted detailed instructions on how bureaucrats were to process applications. These rules, published in full after the breakdown of the GDR, explained which bureaucracies were responsible for processing applications, the information they had to compile, and the criteria for whether to forward the application to higher-level agencies.¹⁰⁸

The first contacts between applicants and the bureaucracy were the local councils (Raete der Staedte und Kreise), the executive bodies of the GDR's third administrative tier. After the council received a written application, applicants had to justify themselves in person. If the council's division for “local matters” saw some likelihood of approval, they sent the application to the local Stasi unit for review. From there, it was sent to each of the following agencies: the District Coordination Group (BKG), the Central Coordination Group (ZKG), and the national-level deputy heads of the Stasi, who had the final authority on the matter.¹⁰⁹ Successful applications after 1983 also needed to be approved individually by Stasi head Erich Mielke, the GDR's second-most powerful person, illustrating the importance of emigration decisions to the regime.¹¹⁰

The bureaucracies often stalled the application process by requiring “numerous, repetitive office visits, followed by unclear answers,”¹¹¹ and demanding endless documentation and permissions from other bureaucratic offices.¹¹² In many cases, applicants were tacitly denied by simply leaving the application open. When officials did make decisions, they usually communicated the decision in person without giving reasons for it. Rejections were timed to avoid important national holidays in case applicants reacted poorly to the news. For successful applicants, local councils helped facilitate their exit.

Sample of East German Emigration Applications

East German Archives

The East German regime is infamous for its copious records on its citizens. By the late 1980s, the Stasi's files covered 125 miles of shelf space, each mile with 17 million sheets of paper.¹¹³ Although not quite as large an operation as the Stasi, local civilian authorities also maintained extensive paper records of emigration applications and their outcomes. We examine records from the civilian archives since they received the initial applications and performed the first filtering, allowing us to include applications that were never sent to the national level.¹¹⁴

We drew our sample from a de-identified list of all application files in the Berlin local civilian archive from 1980 to 1989. We limited the search to this period because data collected by the state had been standardized by then, and this was a period of heightened importance for emigration. This archive encompasses every East Berlin resident who applied for emigration to a nonsocialist foreign country under the GDR.¹¹⁵ To collect data, we had German-speaking research assistants visit the archives, examine each sampled file, and hand-code a set of variables. Given the lack of electronic records and the time needed to code each file, we limited our sample to a single local archive to maximize comparability. Berlin was the ideal choice as the most significant urban center in East Germany and the source of the most politically sensitive emigration. Although we cannot compare decisions about rural versus urban populations, for example, we get a picture of regime strategy when the stakes were highest.

The German government has stringent privacy laws concerning East German records due to the highly sensitive data collected. However, we were given special permission to examine files. Our data collection was limited in two ways to ensure confidentiality. First, much of the data was binned, such as recording an applicant's age range or broad work sector. Second, we could not make copies or take any pictures of the documents, limiting our data collection to what was feasible for research assistants physically present at the archive.

Sampling Procedure

The de-identified list of applications included 27,856 files that were open at some point from 1980 to 1989 and that have been processed by Berlin's local civilian archive. The file names indicated, with reasonable accuracy, the years and outcomes for each applicant, although some of the files we examined were not application files.

We drew a random sample of 600 files from this list, of which 540 were application files. We stratified our random sample over the number of open applications each year and the decisions made each year across the universe of available applications. For instance, 1984 represents 12 percent of open applications in the period, and thus we chose files so that 12 percent were open in 1984. In the full set of applications open in 1984, 62 percent were rejected, withdrawn, or remained open, and 38 percent were approved. We drew our 1984 subsample to also reflect this balance.¹¹⁶

An Overview of the Data

Our procedure produced a random sample of 540 emigration application files, with just under half including multiple applications. Each file contained the entire application history for a given applicant. Although all were open at some point in 1980–89, some individuals first applied for emigration as far back as 1972. In the empirical tests, we begin by testing the outcomes of individuals’ first applications, a comparable sample of distinct individuals unaffected by prior regime reactions. We also take advantage of the rich temporal data to construct two panel versions of our sample: one testing each application (1,021 total applications) and one testing each applicant-year with an open application (1,651 total observations). Summary statistics are shown in Table 1.

TABLE 1. Summary Statistics

Note: The table shows summary statistics for first applications and all applications.

Figure 4 summarizes the timelines of the application files in our sample. For each individual, the figure shows the range of years from the first application to the file's closure (stopping at 1989). Individuals approved for emigration before November 1989 are shown in a darker color. We see a mix of outcomes here, including approvals (especially for those who first applied in 1981–84), individuals who gave up, and applications still open at the regime's collapse.

FIGURE 4. Lifespan of Emigration Application Files

Emigration approval

Our primary interest is predicting which applications the East German regime approved. To code this, we consider anything but formal approval to be rejection, including when individuals withdrew their application (often under pressure). Applications still open at the regime's effective end in November 1989 are coded as rejected to reflect East Germany's frequent practice of keeping applications open indefinitely rather than formally denying them. Emigration approval is 1 for 29 percent of first applications and 35 percent of all applications.¹¹⁷ Among first applications, 34 percent were formally rejected, and the rest were withdrawn or not decided.

Figure 5 provides more detail on approval rates over time. The darker line shows the approval rates for first applications by year of application, with the bars displaying the distribution over time of first applications. The lighter line shows the annual success rate of all applications open in the given year, not just first applications. The temporal patterns fit our expectations from the qualitative analysis. Approval rates were low before 1980, then gradually rose and spiked in 1984. The GDR's leniency that year invited a torrent of new applications, leading the regime to dial back its approval rates. Finally, facing increasing domestic pressure, the government returned to high rates of approval from 1988–89, but primarily for older applications.

FIGURE 5. Applications and Approval Rates by Year

Punishment

Figure 6 shows the variation over time of punishment in response to emigration applications, our other dependent variable of interest. We focus on two indicators of punishment: imprisonment and the initiation of surveillance. Only responses in reaction to the application are counted. Imprisonment was most common early in this period, when emigration applications were formally illegal, and was rare by the early 1980s. Surveillance was far more likely throughout our sample period. By itself, surveillance can be considered punishment, since it frequently included lengthy interrogations and intrusive interference in jobs and social relationships. However, surveillance is best considered a proxy for the state scrutiny that accompanies political, economic, and social punishment. Among first applicants, in all but one case, imprisoned applicants were first placed under surveillance. In total, 58.7 percent of first applicants were imprisoned or surveilled.

FIGURE 6. GDR Punishments for Emigration Applications

Opposition

Our primary explanatory variable of interest is whether applicants expressed active opposition to the regime. We use the regime's record, which is appropriate since we are testing the regime's reaction. We count only active opposition in our coding since this is more threatening and less likely to be overlooked by the regime. However, we separately examine private expressions of opposition revealed during the application process. Active forms of opposition include membership in anti-regime groups; protests; contacting a Western politician, media source, or embassy; quitting a job or ending party membership for political reasons; and contacting GDR officials to express opposition.

We test three versions of this coding. The first is a simple binary coding of any active opposition. The second is a three-valued coding that distinguishes low- and high-threat forms of active opposition. High-threat opposition includes opposition membership, protest, and contact with Western organizations, since these activities are the most public and conducive to collective organization. The third is the number of different forms of active opposition recorded for the applicant. Figure 7 shows the fraction of applicants each year recorded as engaging in any opposition and high-threat opposition activities. Roughly 10 to 30 percent of applicants each year expressed active opposition.

FIGURE 7. Opposition Activity by Emigration Applicants

Other variables

The application sample contains a rich array of data beyond these main variables, some of which we summarize in Table 1. Among applicants, 41 percent were women, and the average age was about thirty-four.¹¹⁸ We also include a binary variable, elderly, coded 1 for those over sixty (about 7% of first applicants), to evaluate whether those at or near retirement age are allowed to exit at higher rates.

We use occupational class, education, and industrial-background measures to capture each applicant's economic importance. Occupational class takes four values: 0, unemployed; 1, manual labor; 2, semi-skilled/skilled labor, and 3, professional. We test this as an indicator of economic value and difficulty of replacement. About 74 percent of applicants fall into the middle two categories, with just over 5 percent rated as professionals. Education ranges from 0 to 5, indicating less than secondary school (12%) up to a university degree (13%).

We also recorded applicants’ reasons for wanting to emigrate (with multiple justifications allowed). Family considerations (including family reunification, cross-border marriages, and care for sick relatives) were the most common, appearing in 59 percent of first applications. Just under a third of applicants cited economic grievances, most commonly dissatisfaction with work or housing. About 23 percent indicated ideological opposition to socialism, while roughly the same number complained about a lack of freedom. The most frequently referenced restriction on freedom was the inability to travel, with smaller numbers citing repression of sexuality or forced military service.

Also of interest is the bureaucracy's analysis of the applicant's reasons for applying, which does not always match the applicant's stated reasons. For instance, among applicants citing ideological opposition, the bureaucracy disputed that this was genuine about a quarter of the time. Conversely, for about 12 percent of applicants not citing ideological opposition, the bureaucrat nevertheless recorded it as a reason. In total, for just under a third of first applications, the bureaucracy disagreed regarding one of the four categories just mentioned (family, economic, ideology, and freedom).

Sample Selection and Unobserved Traits

Before turning to the empirical results, we consider the implications of the self-selected nature of our sample. We only have data on East Germans who applied to emigrate, and it is well known that nonrandom sample selection can cause bias. Specifically, selection bias is a form of endogeneity in which an omitted variable predicts selection into the sample and the dependent variable.

However, some features of the GDR emigration process mitigate against omitted-variable bias. After local agents interviewed applicants using a standardized set of questions,¹¹⁹ they summarized the information, then forwarded the files to higher-ups.¹²⁰ The decisions we are drawing inferences from were made by these high-level agents using these interview summaries, often combined with Stasi intelligence reports. These packaged files were then shared with local agencies and updated as decisions were made, ultimately forming the application files we use for our data. Since we code and test all of the significant elements of these files across our robustness checks, there is little opportunity for omitted variables to interfere in the decision process. Essentially, we can think of the local agents as purging the files of subjective and extraneous omitted variables in their summaries, giving us the full range of information available to the decision makers.¹²¹ Omitted characteristics of applicants were also unlikely to influence the process in other ways. Given the GDR's rigid bureaucracy, applicants had little room to work the system (outside of repeatedly applying) or leverage personal connections.

A related issue is that selection can make the sample unrepresentative of the wider population. But this is also not a major concern since our population of interest is would-be emigrants. Whether a regime would grant exit to a citizen who would never apply is less important. Nevertheless, it's worth noting that our sample is not too dissimilar from the general adult population. Although it skews younger and more male (similar to migrant populations in general), an almost identical percentage of our sample attained a college degree (21.9%) as in all of East Germany (20.6%). Similarly, about the same percentage of working-age adults in our sample were employed (84.7%) as in East Germany (81.4%) and East Berlin (82.0%).¹²² This comparability adds further reassurance that selection is not a severe problem.

Empirical Results

Emigration Approval

We begin our empirical testing of emigration approval with the sample of first applications, ensuring that there is one observation per individual and no prior regime reaction to their application. Table 2 presents a summary of the results. In the first three specifications, we use logit to predict emigration approval from measures of opposition, controlling for the year of application, gender, age, elderly, occupational class, and the number of application pages. The pages variable, which ranges from 1 to 400, captures regime attention on the applicant and accounts for the possibility that greater scrutiny reveals opposition activity. The three models vary by the measure of opposition. Models 4 to 6 repeat this pattern with additional controls for whether the applicant justified their application with economic, family, ideological, or freedom reasons (with multiple reasons allowed).

TABLE 2. Models Predicting GDR Emigration Approval

Notes: The table displays logit models predicting the East German government's approval of emigration applications. The sample is respondents’ first recorded applications. t-values (based on robust standard errors) are shown in parentheses.

* p < .05; ** p < .01; *** p < .001.

As predicted by Hypothesis 1, opposition is positive and significant for emigration approval in every model (using robust standard errors). For each opposition measure, Figure 8 shows the large magnitude of effect from a one-unit change in opposition on the likelihood of emigration approval.¹²³ For the binary measure, opposition increases the chance of approval by 11.9 percent on average. In relative terms, this is about a 50 percent increase in likelihood. Similar effect magnitudes are seen for opposition (threat) and opposition (number), but these measures have larger ranges. Moving from 0 to 2 on opposition (threat) increases the chance of approval by 20 percent, or about 77 percent in relative terms.¹²⁴

FIGURE 8. Estimated Effects of Opposition on Approval and Punishment

Findings for the economic variables in Table 2 also confirm our expectations. A higher occupational class makes emigration approval less likely, as predicted by Hypothesis 3. The effect is substantial—shifting from unemployed to professional cuts an applicant's chances of approval by more than half. In contrast, the elderly are more likely to be allowed exit, as also predicted by Hypothesis 3. According to model 1, non-elderly have about a one-in-four chance of approval, while the elderly have just above a 60 percent chance. In tandem, these findings show a clear picture of the regime maintaining its grip on the most economically productive citizens and ushering retirees out.

Regarding other variables, women are slightly more likely to be approved, although this is not robust.¹²⁵ Age does not matter outside of elderly status. The year of application is not predictive. Perhaps surprisingly, the applicant's stated reasons for emigration are mostly unrelated to approval, including a desire for family reunification. As expected, private ideological opposition is not predictive after controlling for active opposition since quiet dissatisfaction presents little threat. Economic justifications reduce the likelihood of approval, which aligns with our theory because these are precisely the individuals who can be bribed to withdraw their applications.¹²⁶

We tested similar models in two panel versions of our sample, with results summarized in Table 3. For models 1 to 3, the panel includes a separate observation for every year the individual has an open application. emigration approval is coded as 0 until the year of approval (if ever). We control for the year of observation and the number of years since the first application. For models 4 to 6, we use a panel with a separate observation for each application. We control for the year of application and the application number.

TABLE 3. Panel Models Predicting GDR Emigration Approval

Notes: The table displays panel logit models predicting the East German government's approval of emigration applications. The first three models use a respondent-year panel, whereas the next three use a respondent-application panel. Models 5 and 6 limit the sample to respondents’ second or higher applications and third or higher. t-values (based on robust standard errors) are shown in parentheses. *p < .05; **p < .01; ***p < .001.

Opposition (Binary) is again positive and significant (now at the 0.001 level in every model), and the same holds for the other two measures of opposition (see online appendix). Moving beyond solely replicating Table 2, we take advantage of the panel structure to examine how opposition's influence varies over time and application number. In the year panel, model 2 tests whether opposition's effect varies by year, and model 3 tests whether it varies by the applicant's total years of applying.¹²⁷ Surprisingly, we find no variation.

For the application panel, we compare opposition's effect in the entire sample to the effects in only the second or later applications and only the third or later. We find the effect is considerably larger in later applications.¹²⁸ Figure 9 compares the marginal effect sizes of opposition in first applications versus second or more, third or more, and fourth or more. The effect size nearly triples across the categories. This result illustrates the logic of the screening mechanism—it is above all the regime opponents who are determined to leave, applying again and again, that the regime is most intent on releasing. Notably, opposition does not increase its effect with duration of application, only the number of the application. Whereas the former might imply a willingness to wait patiently, the latter reveals a dogged persistence.

FIGURE 9. Opposition's Effect on Emigration Approval by Application Number

Punishment

To make the targeted safety valve a viable strategy, the regime also needs to punish opponents for applying to emigrate. In Table 4, we show results predicting punishment in response to applications. We use the same set of predictors from models 1 to 3 in Table 2, again varying across three measures of opposition. We first predict a binary variable coded 1 if the regime reacted with either surveillance or imprisonment (using logit). Next, we predict a three-valued dependent variable coded 2 for imprisonment, 1 for surveillance only, and 0 for neither (using ordered logit). This takes into account the more severe punishment of imprisonment. Note that we test these outcomes for only the first application. Extending the models to a panel structure is problematic since punishment for first applications directly affects the ability to apply later (e.g., during imprisonment), plus the sentences’ often long durations makes it infeasible to distinguish punishment decisions across applications.

TABLE 4. Models Predicting GDR Punishment for Applying to Emigrate

Notes: The table displays models predicting the East German government's punishments in response to emigration applications. The sample is respondents’ first recorded applications. The first three models use a logit to predict whether the respondent faced surveillance or imprisonment (as a binary variable). The last three models use an ordered logit to predict an ordinal coding of punishment (with imprisonment coded as most severe). t-values (based on robust standard errors) are shown in parentheses. * p < .05; ** p < .01; *** p < .001.

As with emigration approval, opposition is positive and significant for punishment in every model, confirming Hypothesis 2. The effect sizes from models 1 to 3 are shown in Figure 8 alongside the emigration approval results for comparison. For model 1, opposition increases the chance of punishment by 15.5 percent. Shifting from no to high-threat opposition in model 2 increases the likelihood of punishment by 23.3 percent. The results are consistent when using the ordinal measure and controlling for the applicants’ stated reasons for applying.

Looking at the remaining variables, we see that the elderly are much less likely to be punished. This is expected since the regime supported the elderly leaving for economic reasons and faced no concerns about citizens selecting into this category. Occupational class is also significantly negative for punishment, with a shift from unemployed to professional lowering the estimated chance of punishment by 18.2 percent. Unlike the elderly, higher-skilled East Germans faced more restrictions from leaving. This suggests that the regime was intent on keeping professionals at home and out of active punishment to ensure they could remain productive.

Robustness Checks

Table 5 summarizes several robustness checks of our results. For each modification (listed at left), we show the coefficient on opposition (binary) from four different models. The first three predict emigration approval in the first-application model (adapting model 1 of Table 2), the year panel (adapting model 1 of Table 3), and the application panel (adapting Model 4 of Table 3). The last model predicts surveil/imprison (binary) for first applications (adapting model 1 of Table 4). Each coefficient shown is from a separate logit model.

TABLE 5. Robustness Checks for Emigration Approval and Punishment

Notes: The table displays robustness checks for the estimated effect of Opposition (Binary) on emigration approval and punishment. The variations are listed at left. Results are shown for a sample of first applications for both outcomes. For the approval outcome, we also use panels constructed from all applicant-years and all applications. t-values (based on robust standard errors) are shown in parentheses. *p < .05; **p < .01; ***p < .001.

The first three checks address concerns that unusual regime dynamics might confound our results. In particular, the regime faced a severe crisis throughout 1989, so decisions in that year might not reflect a more general strategy. The same goes for the GDR's liberalizing experiment of 1984. Yet results are largely consistent if these years are removed; the lone exception is the prediction of punishment without 1984, which remains of a similar magnitude but just misses significance. The results are also robust to including year fixed effects.

We then include additional control variables. First, we control for whether the applicant submitted an official family application. Second, we control for two binary variables: whether the applicant had close family in West Germany or East Germany, respectively. The latter accounts for concerns that emigrants would remain in communication with their family members. Third, we control for a six-valued measure of education. Fourth, we control for whether the applicant works in government (including the military), social services (including education and health), or primary production (agriculture and manufacturing). The results on opposition are highly robust to these alternate controls.

None of the additional controls is robustly significant (see online appendix), although having East German family is significantly negative for approval in two samples. The lack of findings for education and industry are surprising and seemingly in tension with Hypothesis 3 and the results for occupational class. It suggests that the regime cared about holding on to professionals in all economic areas and did not focus on their education net of current occupational status. In other tests, we found null results (and no substantive change to our opposition findings) from married status, number of children, debt status, and previous attempts at illegal emigration (see the online appendix).

Conclusion

Autocratic leaders—from the kings of feudal Europe to the Chinese Communist Party today—have long aspired to control exit. They continue to use a combination of exit visas, no-travel lists, passport controls, harassment, and bilateral labor agreements to allow some to exit while forcing others to stay. In this paper, we argue that autocrats craft these selective emigration policies to bolster their regime, strategically removing their political opponents to produce more quiescent populations while resisting the loss of productive citizens.

We tested our argument using novel exit application data from East Germany's state archives, the first individual-level test of an autocracy's choices on emigration. Applicants who expressed active opposition to the regime were more likely to be granted exit. At the same time, the regime punished these opponents to deter feigned opposition and screen out all but the most determined dissidents. The regime also considered the economic costs of letting applicants emigrate, with the elderly more likely to secure exit and professionals less likely.

We have argued that selective emigration is a tool widely used by autocrats to stay in power. The targeted safety valve, in particular, appears in politically significant cases like Cuba's Mariel boatlift, the Soviet Union in the 1970s, and modern China and Russia. Scholars should examine how these autocracies have used the targeted safety valve and how the technique has evolved over time.

This raises questions for future research around the strategic mix of co-optation, repression, and emigration policy. State capacity likely affects the adoption and effectiveness of selective emigration policies. When a state cannot control its borders, discontented citizens can simply leave, although such regimes can still leverage many citizens’ preference for legal emigration by controlling passport access. Further, dictatorships with weaker administrative resources may need to abandon individual-level screening in favor of group-level targeting of exit opportunities. Autocrats anticipating bureaucratic noncompliance may also opt for more concrete rules or blanket policies that leave less room for discretion and bribery, or forego strategic emigration policy altogether.

We examined a case where emigrants had easy access to entry and citizenship in a wealthy neighbor. With immigration restrictions rising in the global North,¹²⁹ there are fewer opportunities for easy exit. How often do autocratic sending countries act to provide these opportunities through bilateral migration agreements? Finally, scholars have argued that the cost of repression has increased in the last few decades.¹³⁰ Has this led to a shift toward exit as a strategy to stay in power?

Although East Germany is long gone, its unusually extensive records for its time help us understand how autocracies use selective exit. Technological advances are increasingly enabling state monitoring of citizens without the extensive surveillance apparatus of East Germany. When citizens volunteer their movements, personal networks, and opinions to social media companies, it becomes easier for states to strategically determine who should receive a passport or even who can board a plane. We expect regimes to continue using such data to select out those who might cause trouble, with far-reaching implications for the stability of autocracies and the character of the world's migrant communities.